Slides are here:

• Week 6 Lecture

To prepare for this class:

• Study for your midterm!

A. Please read the following:

• from PDQ Statistics, pages 67-71 “Logistic Regression”
• from PDQ Statistics, re-read pages 56-60, with special focus on choosing model variables

B. Please consider the following questions for discussion in class:

1. What are good reasons for and against dichotomizing or categorizing an INDEPENDENT continuous variable in regression analyses?
2. What are good reasons for and against dichotomizing a DEPENDENT continuous variable in regression analyses?
3. What tools can we use to evaluate whether one regression model is better than another regression model for the same dependent variable?

C. Get your radon dataset ready for analysis

• Create a new variable called Over100 that has a value of 1 for all MainRadon concentrations >100 Bq/m3 and a value of 0 for all MainRadon concentrations <= 100 Bq/m3
• Ensure that you are using the data update that includes the new variable TectonicBelt so that you can follow along with my analyses

Objectives of this class:

A. Exploratory data analysis for a binary dependent variable

• Mosaic plots
• Cross tabulations
• Chi-squared test of association

B. Preparation for logistic regression

• Review of the odds
• Review of the odds ratio
• The logit

C. Logistic regression with categorical and continuous variables

• Interpretation of the intercept
• Interpretation of the other coefficients

D. More on multiple linear regression

• Comparing different multiple logistic regression models
• Comparing different multiple linear regression models